Bias computation

The visibility amplitude bias (there is also a closure phase bias, but this one is nominally corrected by the raw data averaging pipeline) is related to non-Poisson detector statistics due to after-pulsing, and to fixed pattern noise due to non-linearities in the modulation stroke. We determine the bias using so-called incoherent scans, which record data even though the delay lines are deliberately set away from the white light fringe. The fringe amplitude thus recorded is the bias, which is a function of the total photon count rate.

OYSTER keeps track of these incoherent scans, and the bias widget procedure described below only processes those scans.

Before averaging the data for the bias computation, we need to set the background counts to zero. Type at the OYSTER prompt:


Default Backgroundtable created.
Default Background data set.


Averaging rates and squared visibilities...
% Compiled module: MEAN.
% Compiled module: MOMENT.
Averaging triple products...
Averaging delays...
Finished astrometry computations.
Finished visibility estimation.
Finished averaging.

We now use Calibrate|SYSTEM to bring up a special plot widget for system calibrations. It is shown here. (The Bias button and Points directive are clickable fields for more information.)

For our data on Eta Virginis and its calibrator, FKV0445, which are not very bright stars for NPOI, it is best to select all stars except the brightest ones, i.e. BSC1948, FKV0210, FKV0251, and FKV0291, for the purpose of plotting the visibility amplitude of the incoherent scans versus the photon rate. A linear fit to the data plotted in log-log form is performed and the results are stored in GenConfig.Bias (overwriting any previous results). Therefore, if bias corrections are desired for any combination of spectrometer, channel, and baseline, the data of this combination has to be plotted.

In addition, the bias coefficients depend on the number of beams falling on the detector. These different configurations can be selected using the binary fields listed in the Points selection. Fit coefficients are only stored if such a sub-array configuration is selected to make sure they were derived with a well defined setup.

With the data selection as shown above, plot bias fits for all 16 channels of baseline 1 of outputbeam 1, which should look like this:

There is no need for these exercises to plot in such detail all the other fits, so select all baselines and channels and plot the results for the two spectrometers and both configurations. Even though the scale of the plots will prevent you from studying the exact fit results, only by plotting them will the fits be perfomed and stored in the appropriate system calibration tables.

After this has been done, reload the background data (remember that we had to set the background rates to zero for the bias computation).


Backgroundtable created.
Background data loaded.